Support Effect on Stereoselectivities of Vinylogous Mukaiyama

Oct 14, 2013 - Subhrajit Rout , Arko Das , Vinod K. Singh. Chemical Communications 2017 ... Robert A. Feldman , José M. Fraile. Catalysis Communicati...
0 downloads 0 Views 1MB Size
Research Article pubs.acs.org/acscatalysis

Support Effect on Stereoselectivities of Vinylogous Mukaiyama− Michael Reactions Catalyzed by Immobilized Chiral Copper Complexes José M. Fraile,* Nuria García, and Clara I. Herrerías Instituto de Síntesis Química y Catálisis Homogénea (ISQCH), CSIC-Universidad de Zaragoza, C/Pedro Cerbuna, 12, 50009-Zaragoza, Spain S Supporting Information *

ABSTRACT: Chiral bis(oxazoline)- and azabis(oxazoline)-copper complexes, used as homogeneous catalysts or immobilized onto laponite, are able to catalyze vinylogous Mukaiyama−Michael reactions between 2(trimethylsilyloxy)furan and several electron-deficient alkenes. A study of the support effect has been conducted and different changes on the diastereoselectivities and enantioselectivities has been observed. The behavior of the catalyst is different, depending on the structure of the substrate (Michael acceptor). When diethyl benzylidenemalonate was used, the major diastereomer was that with syn configuration, but the homogeneous and the heterogeneous catalysts lead to opposite enantiomers (−80% ee in solution and 38% ee in the heterogeneous phase). This change represents a support effect of ∼1.8 kcal/mol. With N-(E)-but-2-enoyloxazolidinone, the most relevant change is in the diastereomer preference. In solution, the anti isomer is the major one (anti/syn = 98/2); however, in contrast, syn isomer is preferred with the immobilized catalyst (anti/syn = 19/81). This syn preference has not been previously reported in the literature. Finally, in the case of α,β-unsaturated ketones, the homogeneous catalysts are not able to induce enantioselectivity, whereas the immobilized ones lead to moderate values (up to 70%), similar to those values described in the literature with organocatalysts. KEYWORDS: asymmetric catalysis, bis(oxazolines), copper, supported catalyst, support effect



reaction parameters such as metal, solvent, or additives.10 In contrast, only a handful of examples dealing with immobilized catalysts have been reported. Some examples described on silica11 have been explained by the change in the coordinating ability of the anion by hydrogen-bonding with the surface silanols,12 which is a change that is also produced in the case of supported ionic liquid phases.13 The mesoporous materials have been considered as suitable supports for chiral catalysts, as the regular pore system would restrict the conformational freedom of the catalysts and limit the possible pathways for the attack of reagents14 in a sort of confinement effect, invoked also in the case of catalysts supported on zeolites.15 Another type of support for the immobilization of chiral catalysts are the clays, which are layered silicates that have been recognized for a long time as interesting catalysts and supports for organic synthesis.16−18 Our group described, for the first time, the effect of the regular flat surface of clays on the stereoselectivity of copper-catalyzed reactions when the catalyst had been immobilized by electrostatic interactions.9 In the case of cyclopropanation, even the nonchiral catalyst had some influence on diastereoselectivity,19 and the chiral catalysts with bis(oxazoline)

INTRODUCTION Enantioselective catalysis is, in theory, the most interesting method to prepare organic compounds in enantiopure form. In contrast with this idea, and apart from the cases of enzymatic kinetic resolutions, the examples of industrial applications of enantioselective catalysts are rather scarce and mostly concentrated in a short number of hydrogenation reactions.1 One of the reasons adduced for this situation is the high cost and low productivity of most of the enantioselective metalbased catalysts.2 One method to improve their productivity is the immobilization on solid supports,3−6 which, in principle, should allow the recovery and reuse of the expensive enantioselective catalyst, or even permit its use in continuousflow reactors.7 However, immobilization has an additional cost, mainly in the most popular covalent method, because of the required supplementary substitution on the ligand, which is also a source of unexpected effects on enantioselectivity. The use of noncovalent strategies of immobilization8 should help to minimize its cost impact, since the same homogeneous catalysts can be supported and the immobilization procedure is simple and efficient. Moreover, additional advantages should be obtained to compensate the preparation effort, even if not so hard, of the immobilized catalyst (for example, modification of the stereoselectivities of the reaction9). A good number of examples have been described in which stereoselectivities in homogeneous catalysis are reversed by modifications in © 2013 American Chemical Society

Received: August 28, 2013 Revised: September 27, 2013 Published: October 14, 2013 2710

dx.doi.org/10.1021/cs400743n | ACS Catal. 2013, 3, 2710−2718

ACS Catalysis

Research Article

Figure 1. Ligands used in this work and an idealized view of the supported catalysts.

ligands were able to reverse both diastereoselectivity and enantioselectivity.20 Since then this effect has been studied indepth by using different supports,21 ligand substituents,22 unsymmetrically substituted bis(oxazolines),22,23 monooxazoline ligands,24−26 and substrates and diazocompounds.22 Unfortunately, the system was revealed to be rather unpredictable to get a full picture of the surface effect, probably because of its high conformational flexibility, and different effects on selectivities (enhancement, decrease, or reversal) were obtained. In the meantime, effects on other reactions were also observed. In the case of carbene insertions into C−H bonds of cyclic ethers, stereoselectivities and also chemoselectivity were improved.27,28 An impressive effect, with reversal of diastereoselectivity and great enhancement of enantioselectivity, was observed in the case of the vinylogous Mukaiyama aldol reaction using 2-(trimethylsilyloxy)furan.29 This enolsilane is very useful from a synthetic point of view, because it leads to γ-butenolides30 that are present in a large number of natural products. The related vinylogous Mukaiyama−Michael reaction31,32 has been less explored, with only a limited number of examples of the enantioselective version catalyzed by bis(oxazoline)-metal complexes described in the literature.33 In this case, it is also important to highlight that, to our knowledge, there are no reported examples of using heterogeneous catalysis to promote this reaction. In this paper, we present an extensive study of the surface effect on enantioselective vinylogous Mukaiyama−Michael reactions between 2-(trimethylsilyloxy)furan and several electron-deficient alkenes, showing how the structure of the substrate (Michael acceptor) conditions the results of both diastereoselectivity and enantioselectivity. Cu(II) complexes of different bis(oxazoline) (box)33 and azabis(oxazoline) (azabox)34,35 ligands (Figure 1) were tested as catalysts. Laponite, which is a synthetic layered magnesio-silicate,36 was used as a support for the immobilized catalysts. Immobilization was carried out via the exchange of some of the Na+ cations by boxCu(OTf)2 complex in methanol,37 leading to a more disordered material (Figure 1).38 The catalysts were characterized by copper and elemental analysis (see the Supporting Information) and by infrared (IR) spectroscopy,22,37,38 to confirm the structural integrity of the complex after cation exchange.

Scheme 1. Reaction between Diethyl Benzylidenemalonate and 2-(Trimethylsilyloxy)furan

Table 1. Results of the Reaction between Diethyl Benzylidenemalonate and 2-(Trimethylsilyloxy)furana Homogeneous

Immobilized

ligand

yield (%)

syn/anti

%ee synb

%ee antib

yield (%)

syn/anti

%ee synb

%ee antib

1a 1b 1c 1d 2b 2c

100 100 100 100 100 100

94/6 79/21 90/10 94/6 77/23 86/14

−52 −23 −83 −80 15 −62

27 −26 63 2 −50 50

100 86 100 98 71 65

79/21 67/33 78/22 83/17 75/25 74/26

26 4 15 38 −3 2

−6 1 21 17 −2 7

a

Reaction conditions: 1 mmol of diethyl benzylidenemalonate (226 μL), 1.5 mmol of HFIP (159 μL), catalyst (0.10 mmol of homogeneous and 0.15 mmol of heterogeneous), 5 mL of anhydrous toluene, slow addition (5 h) of 2 mmol of 2-(trimethylsilyloxy)furan (347 μL) in 10 mL of anhydrous toluene at room temperature (rt). Reaction time after addition: 5 h in the homogeneous phase and 24 h in the heterogeneous phase. bConsidered positive when the major product is that of lower retention time.

the chiral ligand reduces the diastereoselectivity, slightly in the case of ligands with aromatic substituents (1a and 1d) and more significantly in the case of ligands with aliphatic substituents, and mainly with tBu (1b and 2b), leading to the lowest diastereoselectivity values (around 78/22). With respect to enantioselectivity, box ligands are slightly better than the analogous azabox ones, whereas the worse results are obtained with the tert-butyl substituted ligands. Under these conditions (room temperature (rt), toluene solvent, 1,1,1,3,3,3-hexafluoroisopropanol (HFIP) additive), 1c and 1d (ee over 80%) are the best ligands for the homogeneous reaction, despite the better results described for 1b at low temperature. The enantioselectivity in the minor anti isomer did not show any correlation with that of the major syn isomer, with 63% ee as the best value obtained with 1c. The fluorinated alcohol was necessary to accelerate the reaction,39 but its role in the control of the enantiomeric excess has not been described. Several fluorinated alcohols (2,2,2trifluoroethanol as a primary alcohol; 1,1,1,3,3,3-hexafluoro-2-



RESULTS AND DISCUSSION Reaction with Diethyl Benzylidenemalonate. The first Michael acceptor considered was diethyl benzylidenemalonate (Scheme 1), which had been used in nonvinylogous Mukaiyama−Michael reactions catalyzed by the same type of complexes.39 The results, both in homogeneous and heterogeneous phase, are collected in Table 1. In the homogeneous phase, syn (unlike) products are the major ones, as it happened in the reaction catalyzed by Cu(OTf)2 (syn/anti = 98:2),40 but in general, the presence of 2711

dx.doi.org/10.1021/cs400743n | ACS Catal. 2013, 3, 2710−2718

ACS Catalysis

Research Article

Figure 2. (Left) Overimposed geometries (according to X-ray diffraction) of the [1a-Cu-benzylidenemalonate](SbF6)2 (green) and [1aCu(H2O)2](OTf)2 (deep blue) complexes. Hydrogen atoms are omitted for the sake of clarity. (Right) Proposed attack route.

(syn/anti = 75:25)40 and the presence of the ligand does not significantly modify this selectivity, with values in the range from 67/33 to 83/17. At the same time, the sense of the asymmetric induction in the major syn isomer was reversed with respect to the homogeneous values and the major enantiomer was that of opposite configuration. A similar effect had been already observed in the case of cyclopropanation catalyzed by bis(oxazoline)-Cu(I) complexes immobilized on laponite, and it had been ascribed to the disposition of the complex with respect to the surface of the support.20,26 In this case, the effect is observed with Cu(II) catalytic complexes, showing its general character. The most remarkable reversal was observed with 1d ligand, from 80% ee in solution to 38% ee of the other enantiomer with the immobilized catalyst. Such a change in enantioselectivity represents a variation of ∼1.77 kcal/mol in the relative energies of the corresponding transition states, which must be due to the surface effect. In fact it can be speculated that the box-Cu-benzylidenemalonate complex, once the two anions have been exchanged by the negative charges of the support, will be placed on the surface to minimize the steric interactions, so probably the less hindered face in solution will be shielded by the surface, explaining, in this way, the reversal in the induction sense. The low enantioselectivity would be the consequence of the shielding effect of the ligand on the lesshindered face. However, this hypothesis, which considers the intermediate as a rigid body, is a simplification, as demonstrated in analogous examples.22,23,25,26 Reaction with N-(E)-but-2-enoyloxazolidinone. The Nacyloxazolidinones are among the most-used Michael acceptors, and specifically the reaction with 2-(trimethylsilyloxy)furan had been described in the literature, using BINOL-lanthanides,45 bis(oxazoline) and pyridinebis(oxazoline) complexes with different metals,46 or binaphthyldiimine−Ni47 chiral catalysts. In our case, the results obtained with N-(E)-but-2-enoyloxazolidinone (Scheme 2) and bis(oxazoline)-copper complexes in solution and in heterogeneous phase are gathered in Table 2. In this case, the major diastereomer has anti relative configuration (syn/anti ratio from 8/92 to 2/98) in the homogeneous reaction, in agreement with the results obtained with other chiral catalysts45−47 and with Cu(OTf)2 or Lap-Cu without chiral ligands.40 The use of the immobilized catalysts

methyl-2-propanol as a tertiary one; 1,3-(2,2,2-trifluoroethoxy)2-propanol41 as a more hindered secondary one) were tested in the reaction catalyzed by 1c-Cu(OTf)2. In all cases, the selectivities were very similar, with a syn/anti ratio between 85:15 and 90:10 and an enantiomeric excess of the syn isomer between 79% and 83%. This seems to indicate that the kinetic effect is due to a change in the rate-determining step, but the participation of fluorinated alcohol does not take place in the stereoselectivity-determining step (furane attack) but in the silane transfer step necessary to close the catalytic cycle. Although crystals of the major syn isomer were obtained, they were not suitable to determine the absolute configuration. Hence, the sense of the asymmetric induction cannot be inferred from experimental data. The structure of the 1a-Cubenzylidenemalonate complex has been described, albeit with SbF6 anions,39 and this structure is very similar to that of the diaquo complex with triflate anions.42 In fact, the overimposition of both structures (Figure 2) indicates that the oxygens are nearly in the same position, slightly distorted from the equatorial plane, and the anions are also in the same position, irrespective from their nature. The only significant variation corresponds to the rotation of the phenyl groups around the bond to the oxazoline ring, probably due to the presence of the benzylidenemalonate. Thus, it can be speculated that the attack would take place through the Re face of the benzylidenemalonate (Figure 2), leading to syn-3S as the major product. However, in the related reaction between benzylidenemalonate and indole catalyzed by 1c-Cu, the attack through the Si face has been proposed to explain the induction sense in nonpolar solvents,43 based on the square pyramidal structure of the analogous [1b-Cu(H2O)2](OTf)2 complex.44 With this model, the major product would be syn-3R. Interestingly, the induction sense in the syn isomers with all the bis(oxazolines) is the same irrespective from the substituent nature, in contrast with the result reported in the nonvinylogous Mukaiyama−Michael reaction on the same substrate.39 In the case of the immobilized catalysts, the diastereoselectivity was generally lower than that obtained in solution with their analogous catalysts. This immobilization effect had been already observed with the nonchiral heterogeneous catalyst 2712

dx.doi.org/10.1021/cs400743n | ACS Catal. 2013, 3, 2710−2718

ACS Catalysis

Research Article

Scheme 2. Reaction between N-(E)-but-2enoyloxazolidinone and 2-(Trimethylsilyloxy)furan

enantioselectivity can be ascribed to a disposition of the complex with the less-hindered face toward the solid, leading to a competition between the shielding effect of the ligand substituent and the surface. The explanation for the reversal of the diastereoselectivity is, by far, more difficult to justify, mainly because the reaction mechanism is not well understood. The enantioselectivity is explained by the disposition of the Michael acceptor in an s-cis conformation, in agreement with X-ray diffraction (XRD) data,48 and the approach of furan by the less-hindered Si face (Figure 3). However, the disposition of furan, with respect to the acceptor, is still a matter of debate. Both open transition states and cyclic ones (similar to the Diels−Alder transition states) have been proposed to explain the stereochemical outcome of Mukaiyama-type reactions of silyloxyfurans.49 Open-chain transition states allow one to explain the results of uncatalyzed Mukaiyama−Michael reactions with quinones,50,51 but they have been also proposed for Sc-catalyzed reactions with unsaturated ketones,52 although in that case, a coordination of furan to Sc has been also envisaged. Diels− Alder-like transition states have been always proposed in both Lewis-acid-catalyzed Mukaiyama aldol53 and Mukaiyama− Michael54 reactions. In the case of copper catalyzed reactions of simple enolsilanes with N-acyloxazolidinones, a mechanism through a hetero-Diels−Alder state has been observed, but in that case the Michael acceptor acts as heterodiene and the enolsilane as dienophile.55,56 When all the possible approaches of furan by the Si face of the Michael acceptor are represented (Figure 3), no clear preference for any of the transition states can be observed. In the case of Diels−Alder-like transition states (TS-DA), the one leading to the major anti diastereomer corresponds to the endo approach, the kinetically favored in a Diels−Alder reaction,57 although it has been described that reactions with furan as a diene are reversible up to −20 °C.58 The presence of HFIP might prevent this reversibility and TS-DA would explain the observed diastereoselectivity. Regarding the open-chain transition states (TS1 and TS2), the antiperiplanar disposition of the double bonds in TS1 has been always proposed as the most favorable approach. TS1-anti has also the advantage over TS1syn that H atoms are in antiperiplanar, minimizing the steric interactions between the substituents. However, TS1-anti places the bulky TMS group toward the catalytic complex

Table 2. Results of the Reaction between N-(E)-but-2enoyloxazolidinone and 2-(Trimethylsilyloxy)furana Homogeneous

Immobilized

ligand

yield (%)

syn/ anti

%ee synb

%ee antic

yield (%)

syn/anti

%ee synb

%ee antic

1a 1b 1c 1d 2b 2c

100 100 100 100 100 100

2/98 7/93 7/93 8/92 7/93 4/96

69 19 23 34 13 21

36 62 16 59 92 26

81 70 100 76 85 81

64/36 35/65 81/19 67/33 35/65 57/43

32 2 0 52 15 9

32 20 52 46 29 28

a 1.5 mmol of HFIP (159 μL), catalyst (0.10 mmol of homogeneous phase and 0.15 mmol of heterogeneous phase), 5 mL of anhydrous toluene, slow addition (5 h) of 2 mmol of 2-(trimethylsilyloxy)furan (347 μL) in 10 mL of anhydrous toluene at rt. Reaction time after addition: 5 h in the homogeneous phase and 24 h in the heterogeneous phase. bsyn-4SR is the major syn isomer. canti-4SS is the major anti isomer.

promotes the formation of the syn isomer in different degrees, depending on the chiral ligand usedfrom syn/anti 35/65 with tert-butyl substituted ligands 1b and 2b up to syn/anti 81/19 with 1c, showing that this result is a combined effect of the support and the ligand. This is a remarkable result since, to the best of our knowledge, none of the catalysts described in the literature promotes the formation of the syn isomer. Unfortunately, enantioselectivities are always lower than those obtained with the homogeneous catalysts, especially for the major syn isomer. The only exception is the result with Indane box 1d, leading to moderate diastereoselectivities (67/33 d.r.) and enantioselectivities (52% ee for syn). Again, the decrease in

Figure 3. Possible approaches of 2-(trimethylsilyloxy)furan to the Si face of coordinated N-(E)-but-2-enoyloxazolidinone. 2713

dx.doi.org/10.1021/cs400743n | ACS Catal. 2013, 3, 2710−2718

ACS Catalysis

Research Article

expected, the homogeneous reactions do not produce any asymmetric induction (ee